The new GPUs feature HBM4 high-bandwidth memory, advanced security capabilities and AMD’s open ROCm software platform, enabling organizations to deploy and scale AI workloads without vendor lock-in.
According to AMD, the MI455X is optimized for large-model training and AI inference, while the MI430X delivers up to 288 TFLOPS of FP64 performance for scientific and HPC applications.
The company said the MI400 Series also supports secure boot, encrypted GPU-to-GPU communication and scalable system architectures, allowing customers to build AI infrastructure for everything from national research centers to hyperscale AI factories.